The hype surrounding content-creating artificial intelligence (AI), embodied in the AI chatbot ChatGPT, has drawn significant attention from the private sector in Latin America.
According to data compiled by Similarweb, it is estimated that there were over 3030n visits to ChatGPT from Latin America by March this year, accounting for about 10% of the visits during this period. Brazil was the Latin American country generating the most access, followed by Colombia and Mexico.
Rodrigo Scotti, co-founder of the Brazilian AI association Abria, believes AI solutions will continue to grow and develop, but some experts warn investors and businesses that local residents and their revenue We caution you to focus on practical use cases that can have a meaningful impact on your
“Advanced Generative AI is a tool like no other. Technology evolves exponentially,” Scotti told BNamericas.
In any case, some experts believe that these new tools will take the debate to another level, such as the prospects for future cognitive development of individuals faced with the massive outsourcing of creative activities and work to machines. I’m thinking of raising it.
Scotti, who is also CEO of AI company Nama, acknowledges that issues such as intellectual property, legal liability for content, and data protection may have to evolve at the same time. But he believes the current public debate is biased and generally unhelpful, saying AI could fill many productivity gaps in Brazil.
“Compared to traditional programming, it’s impossible to be a binary, good guy vs. bad guy. There are still stories of unemployment – we’re all going to live in a dystopian future. Let’s try to understand what is happening, ”he says.
Scotti’s Nama has seen a double-digit increase in AI inquiries from companies since December. Most of these were consultations, but the executive said his company, which has been working on AI solutions based on large-scale language models (LLMs) since 2019, has several projects lined up.
LLM is a fundamental machine learning model that uses deep learning algorithms to process and understand natural human language.
Management software companies are also evaluating solutions for enterprises based on technologies such as ChatGPT.
“The tests that are being carried out are on representative clients from different industries, especially in Latin America, where we have companies based in Argentina,” said Claudia Boeri, CEO of SAP Latin America South. told BNamericas. “SAP’s intention is to be able to generate reproducible use cases in these industries.”
Internally, the German software giant uses AI to speed up and reduce programming time in its software creation labs, and ChatGPT to generate lines of code.
According to Boeri, SAP typically tests software with customers for a year before releasing it to the market. These processes can be automated with AI to reduce these times.
SAP has also started talking to customers about integrating ChatGPT into their enterprise resource planning (ERP) software. The case is still in its very early stages and remains confidential at this time.
Management software company Benner plans to invest R$13 million (US$2.6 million) this year in product innovation, particularly research and development of optimized generative AI tools.
“this month [April], releases the first solution developed with ChatGPT for the Legal and HR fields. Through the first half of 2023, we also plan to make other applications available to other verticals served by Benner, such as health,” said Marcelo Murilo, vice president of innovation and technology at the company, in a statement. says.
Programmers, developers and analysts on Benner’s innovation team are working to create new tools that can be used across the market, including companies not running Benner’s ERP software, the company said.
A recent Deloitte survey found that 7 in 10 Brazilian companies plan to invest in AI this year. Data he collected from February to March from 501 companies with total revenues representing 21% of Brazil’s GDP.
In general, the main front for testing and using advanced AI and natural language processing (NLP) is customer service, especially via chatbots. But that is changing rapidly.
Brazil’s state-owned oil giant Petrobras is testing advanced AI in its supply and procurement operations, BNamericas has learned. These tests are done in partnership with Nvidia and Supermicro along with the development of the supercomputer.
data infrastructure
The massive amount of data and consumption generated by advanced AI models also impacts data centers in terms of available space, connectivity, and energy consumption.
Fast response times of these AI systems require increasingly lower latency, which means better connectivity, but increased server capacity means more power consumption.
“These GPU clusters are superservers serving artificial intelligence and consume more watts and power. US manager Alexandre Alves told BNamericas.
Data centers are projected to consume 10% of the world’s energy in the next five to seven years.
“Historically speaking, we had the internet, cloud computing, internet of things, 5G, and now artificial intelligence. Bitcoin and blockchain also brought excessive energy consumption. We are never in our comfort zone,” says Fernando Ribeiro. Telecom system coordinator at Odata.
“Our real concern right now is connectivity and low latency. When it comes to energy, we monitor this process.”
The right IT infrastructure to support advanced generative AI can prove too costly to maintain on-premises rather than in colocated data centers, Ribeiro adds.
Data center companies must adjust to meet that additional demand. Vertiv’s Alves says:
Training and deploying AI models in data centers also involves consuming large amounts of water to generate electricity to power and cool servers.
According to a study titled “Making AI Less Thirsty: Uncovering and Addressing the Secret Water Footprint of AI Models,” produced by the University of Colorado Riverside and the University of Texas, ChatGPT consumes 500ml of bottled water for every 20-50 conversations. Consume the corresponding amount. questions and answers.
We also found that training GPT-3 in Microsoft’s state-of-the-art US data center directly consumes 700,000 liters of fresh water (equivalent to producing 370 BMW cars or 320 Tesla electric cars). enough).
“As far as I know, physically speaking, there are still no language models that work in Latin America,” Alves said of the advanced AI data location.
“But if these systems are to become a reality for day-to-day search and operations, we need to build a lot more hyperscale.”
With additional report by Leticia Pautasio
